pysr

Automatic parametric modeling with symbolic regression

Data and helper script for the mini Hackathon to learn symbolic models from satellite orbits.

The first part for hackthon for the symbolic model discovery workshop

Repository dedicated to the inference of physical laws from synthetic data using Symbolic Regression (SR) and Genetic Algorithms.

Code and research for a novel Constraint-Interaction Metric M(i,j) in Symbolic Regression. Built to measure synergy and redundancy between search space constraints using grammar-based AST generation. (TIH IITG Summer Research Project)

Experiments with NAI (No-Ansatz Inference), a symbolic-regression system that infers the form of a law from data without an LLM: benchmarks, protocols, frozen binaries, results and reproduction scripts. No source code.

A mathematical relation of the temperature, radius and luminosity with the Absolute Magnitude of a given star is derived using PySR library. Instead of building deep neural networks or complex ML algorithms, PySR simply tries to built mathematical expressions that best describe the relationship between variables in a dataset.

Plasma Data Analysis with PySR

MCP server for symbolic regression — discover formulas from data, simplify expressions, check equivalence

Symbolic regression with PySR: rediscovering Kepler's third law from NASA exoplanets and interpretable formulas vs gradient boosting.

Symbolic regression vs MLP on Kepler’s Third Law: a Law-Discovery Machine (LDM) using PySR. (Has Paper with it)

Using symbolic regression and genetic programming to discover an interpretable, closed-form equation for customer segmentation.

This project was done as Master's Thesis in Institute of Particle Technology within TU Braunschweig. An autonomous LLM agent built using the Model Context Protocol (MCP) that optimizes symbolic regression on real industrial datasets from particle technology research.

Code for "Physics-Anchored Discovery of Trustworthy Pathloss Models": a discovery method whose output is physically valid and structurally deterministic, plus the evaluation tools and full experiment records.

AI4S Agent assessing 14 organic-chemistry propositions with independent computation, layered evidence, and explicit scope boundaries.